Presentation within the framework of the 11th International Conference on the Promotion of Educational Innovation, Larissa, October 17–19, 2025.
This research examines the pedagogical effectiveness of generative Artificial Intelligence tools, such as Gemini AI and MagicSchool, in the teaching of Mathematics to middle school students. By employing targeted prompts and assigning a guiding role to Large Language Models (LLMs), their ability to solve realistic mathematical problems step by step and to foster critical thinking was investigated.
The results indicate that LLMs respond positively, providing structured analysis and encouraging feedback, thereby enhancing mathematical reasoning. However, certain limitations were observed, including inconsistency in prompt application, errors in cases of incomplete data, and linguistic challenges with the Greek vocabulary. In conclusion, while Artificial Intelligence demonstrates significant potential to support teaching, the critical need for the active presence and supervision of an experienced educator is highlighted.